Software entities are more complex for their size than perhaps any other human artifact since in a software system no two parts are alike. Mircea Lungu seeks to understand how we can build tools that augment the human intellect to help it cope better with the inherent complexity of both writing new software and reading software that is already written. His research lays at the intersection of empirical software engineering, programming language design, and data science.
To provide approaches that scale up to the reality of the large amounts of source code that exist in the context of industrial software ecosystems and in open source code repositories, Dr. Lungu is developing hybrid solutions that trade some of the formalism (and soundness) of the traditional static source code analysis for the scalability of data science methods such as information visualization and data mining.
He has recently become interested in investigating the potential that comes with the ubiquitous tracking of user behavior for build evolving models of user knowledge that can later enable the adaptation of software tools to the particular needs of the user in context. To study this, he is currently designing an architecture for an open and intelligent learning software ecosystem. Since in this project, machine learning will play an important role, he is interested in collaborating with the AI experts from within the DSSC.